[Feat] Resume Site Creation

[RichardAnderson]

This pull request introduces a robust retry mechanism for site installations, improves error handling and reporting for installation failures, and refines the behavior of several background jobs. It also enhances the tracking of installation progress and error details on the Site model, and adds various safety checks to the site setup process.

Site Failed Banner

Retry Mechanism & API Enhancements:

• Introduced a new RetrySite action, allowing sites in an installation_failed state to be retried. This is exposed via new API and web routes, with appropriate authorization and validation.

Error Handling & Progress Tracking:

• Enhanced the Site model and resource to include progress_step and last_error fields, enabling more detailed progress tracking and user-friendly error summaries. These fields are set and updated throughout the installation process and surfaced in API responses.
• Improved error handling in CreateJob by setting a UI-safe error summary on failure, avoiding leaking sensitive or verbose exception messages, and resetting progress fields on success.

Job Dispatching Consistency:

• Standardized job queue assignment by moving onQueue('ssh') from dispatch calls to the constructors of CreateJob, DeployJob, ExecuteCommandJob, and RollbackJob, simplifying job dispatching throughout the codebase.

Site Installation Safety & Idempotency:

• Refactored AbstractSiteType to add idempotent checks for user creation, repository cloning, and FPM pool setup, reducing the risk of errors on retries or partial installations. Added helper methods to check for existing users, cloned repositories, and FPM pools.

These changes collectively improve the reliability and user experience of site installations, especially in failure and retry scenarios.

[Copilot]

Pull request overview

Adds a "resume / retry" capability for failed site installations. Sites can now be retried (web + API), the install pipeline records a named progress step and a UI-safe last_error, and the install steps (user creation, repo clone, FPM pool) are made idempotent so retries skip already-completed work. Job onQueue('ssh') calls are moved into the constructors of CreateJob, DeployJob, ExecuteCommandJob, and RollbackJob.

Changes:

• New RetrySite action plus web/API routes, OpenAPI docs, banner UI with confirmation dialog, and a feature test.
• Site gains progress_step and last_error columns; AbstractSiteType::progress() accepts a step label; CreateJob records a sanitized error summary on failure and resets fields on success.
• AbstractSiteType gets userExists / repositoryAlreadyCloned / fpmPoolExists guards (with new Blade probe scripts); nginx/caddy vhost symlinks switch to ln -sfn and create-path drops rm -rf for mkdir -p.

Reviewed changes

Copilot reviewed 36 out of 37 changed files in this pull request and generated 13 comments.

Show a summary per file

File	Description
app/Actions/Site/RetrySite.php	New action validating state and dispatching CreateJob.
app/Actions/Site/{CreateSite,Deploy,ExecuteCommand,Rollback}.php	Drop ->onQueue('ssh') (now in job constructors).
app/Http/Controllers/SiteController.php	Adds web sites.retry route.
app/Http/Controllers/API/SiteController.php	Adds API retry endpoint.
app/Http/Resources/SiteResource.php	Exposes progress_step and last_error.
app/Jobs/Site/CreateJob.php	Sets queue in constructor, resets/sets progress fields, adds safeErrorSummary.
app/Jobs/Site/{DeployJob,ExecuteCommandJob,RollbackJob}.php	Set queue in constructor.
app/Models/Site.php	New progress_step / last_error properties + fillable.
app/SiteTypes/AbstractSiteType.php	New idempotency helpers and progress($percent, $step) signature.
app/SiteTypes/{Wordpress,PHPSite,PHPMyAdmin,PHPBlank,NodeJS,MiseNodeJS,MiseBun,LoadBalancer}.php	Adopt step labels and cloneRepository().
database/migrations/2026_05_16_155350_add_progress_step_and_last_error_to_sites_table.php	Schema migration for new columns.
public/api-docs/openapi/{sites.yaml,schemas/Site.yaml}	Documents retry endpoint and new fields.
resources/js/components/site-banners.tsx	Adds failed-install banner with retry dialog.
resources/js/pages/servers/components/header.tsx	Shows current install step inline and in tooltip.
resources/js/lib/utils.ts	New humanizeStep helper.
resources/js/types/site.d.ts	Adds typing for new fields.
resources/views/ssh/services/webserver/{nginx,caddy}/create-vhost.blade.php	Use ln -sfn with escapeshellarg.
resources/views/ssh/services/webserver/{nginx,caddy}/create-path.blade.php	Replace rm -rf + mkdir with mkdir -p.
resources/views/ssh/site/check-*.blade.php	New SSH probes for user / repo / FPM pool.
tests/Feature/RetrySiteTest.php	Coverage for retry flow, validation, and failure handling.

[RichardAnderson]

Still draft, needs more testing